Making Peace with AI-Generated “Good Enough”

On design, irrelevance and cognitive agency

I always thought that human creativity would be our last frontier in this dynamically changing landscape of AI automation. An always safe, protected area where we all could take refuge when things get out of hand. Something that is deeply human, irreplaceable and genuinely authentic to each one of us.

Although all of this might still be true, the question is whether human creativity is valued enough by us humans. Turns out the biggest risk might not be what AI can do, but what we as society appreciate and value.

—

A few weeks ago, for the first time, I actually began to feel worried about the future of the graphic design profession because of AI. It was during one of my deeper sessions with Claude - specifically Claude Design. Still in beta at the time, it offered generative creation of presentation slides, websites and motion design. Because curiosity is what naturally drives me (and often gets me in trouble), I decided to try it out and see what it can do with properly prepared brand assets. First, I had a co-work session with Claude to prepare the brand kit folder in a way that would guard against potentially wrong (or too loose) interpretations. Then, I uploaded it to Claude Design.

Before my eyes, Claude started crunching on the whole brand design system as if the brand system were its morning cereal and turned it into code somewhere in the background. I could see fragments of its inner chatter that were too fast to grasp, but I was left with the uneasy feeling that it knew something I didn’t. It instantly triggered childhood memories of schoolmates whispering in each other’s ears while looking at me. I felt left out and irrelevant.

They knew something I was not worthy of knowing, much like Claude keeps its inner reasoning to itself. It is very hard not to take it personally. It surfaces bits here and there - enough to make you aware of its reasoning, but not enough to understand it. Like those girls - they were certainly not whispering to avoid being heard.

I had to consciously remind myself that this is a result of a specific user experience designed by the company called Anthropic - not some evil joke from AI. But that feeling of unease remained. Months of developing the brand visual language were processed within a few seconds, translated into a logic I don’t understand. It wasn’t instant, but it was certainly fast enough to make me feel dumb for working on it for so long.

Things are going to be different from now on. My non-designer team members will easily generate new designs from the reusable design system I’ve created. Seems like I just undermined my very existence in the company for good. All the visuals I previously had to derive from the brand system look foolishly trivial and tiresome now. Overwhelmed as I was with all kinds of emotions, I kept observing Claude’s process while holding my breath, as if I were watching a nerve-racking thriller. Is it going to be useful at all? Or will it be another mockery, like AI-generated typefaces or 3D?

Finally, the results presented themselves: a fully functional brand system with presentation slide templates and other useful formats. It looked well organized and promising: the typography styles, the colors, the use of brand photographs - all neat. However, in my career I had seen quite a few projects that started promisingly and then failed to deliver a proper result - so I didn’t trust it just yet. I needed to test it on something challenging. I chose the most annoying task I had done in the previous few weeks: a branded animated video of a countdown clock. This is certainly something you could not generate from a simple text prompt with any of the AI video tools before. Despite my skepticism, I felt that this time it might actually work - because of the amount of input I had already given Claude. And oh yes, Claude did deliver.

A perfectly solid animation showed up in front of my eyes: with falling elements in the background, countdown numbers in the foreground, perfectly on brand, with all the necessary information it was supposed to contain. It was… actually GOOD! Damn…

I had to leave the computer immediately. No, I had to leave the house! I felt a mix of excitement and anxiety, terrible restlessness and the feeling that I couldn’t breathe. I took my dog Lola out and ended up hyperventilating in front of the Institute of Electronics and Photonics. I remember I was, for some reason, trying to calm myself down by looking at a tree. As if the tree would make everything go away. Or maybe it was because a tree is at least something constant, something that does not change much over time.

This was not the shock of being confronted with a new technology demonstration. I have been working closely with AI for many, many years. I got used to the constant rapid updates and new breakthroughs, repeating to myself in a celebratory voice “what a time to be alive!” (nod to the Two Minute Papers YouTube channel). So I had to ask myself: “why the drama?” I am not an endangered species, I don’t depend on graphic design for my income - I do this as research. And even if I did, I know my particular set of skills makes me difficult to replace with AI.

I still don’t have the answer. I speculated it was maybe sheer nausea from the speed with which Claude delivered the result - something like motion sickness in a car. But something tells me that maybe it was disgust at the absolute reduction of a very nuanced human creative work into something as cold as code.

Designing a scalable visual system that is unique and carries the DNA of the brand (a combination of specific sentiment, ideology, vision and visual symbolism native to the company’s target audience) is not an easy task. It’s much more than just deciding on the colors and fonts. If that were the case, all brands would look the same. Giving the brand its look is also about something deeper than just the curation of visual elements. I think I speak for every graphic designer when I say that it requires a human being to tune into the brand’s vision and goals, translate it into an emotion, and subsequently find a matching visual representation for it.

The visual system I created, with all its delicate rules and nuances, had been processed through programming logic and deconstructed into a set of abstracted rules, which I can’t verify because I don’t have access to them. It’s the whispering girls over and over again, now deciding how to (ab)use my creative work so they can be the ones to shine. Yes, I’ll admit it - this is some kind of strange jealousy. It was also an uncanny feeling of losing control and feeling locked out of my own creative process.

It wasn’t the first time I had seen someone butcher my designs. Designers actually encounter this quite often. You design the brand system, make a neat design manual, and deliver assets that someone unskilled will work with later. What happens next is usually anyone’s guess (or in the hands of a self-proclaimed graphic designer, aka a secretary). It hurts to see bad visual outputs from a well-designed brand system. What hurts even more, though, is seeing good visual outputs from a well-designed brand system - made by AI.

Let’s not overhype this. The animation Claude generated was not the most polished animation I had ever seen. But it was GOOD ENOUGH. It was solid. And that scared the shit out of me.

Somewhere between what could already have been my thirtieth shorter inhale and longer exhale, I realized that what I had always believed was protecting me - my creativity - might not be protection at all. Not because AI is more creative than me. But because most clients don’t need the kind of creativity I offer. What most people really need is just “good enough”. And good enough is now accessible to everyone.

The market only needs “good enough”

For years, I have been defending AI as an amazing tool that is here to amplify our creativity, not replace it. Now I feel I should perhaps reframe my thesis: AI cannot replace human creativity, but it can imitate it well, reduce it to good-enough outcomes and decrease the need for human-made content. In this particular case, that would mean fewer mid-level design jobs. But it could be TV shows, books and music too. As long as there is an incentive to keep the price low, nobody would pay a human creator if hiring one isn’t necessary. Humans are expensive. Human creativity is a premium service.

Clients can now do much more themselves, whether by using an existing brand manual as an input for Claude’s workspace session, or asking ChatGPT to design a logo. The work opportunities for designers are shrinking fast with each new AI model update. Companies are implementing AI and redesigning their workflows to be more efficient, which might reduce repetitive tasks but also make junior-level employees suddenly obsolete.

This is a real existential crisis for designers. If you studied for 6-10 years to become one, then struggled in unpaid internships to gain necessary practical experience and woke up to this paradigm shift, you cannot simply switch careers and become an enthusiastic masseur. It genuinely hurts to realize that something you cared so much about and dedicated tons of energy to is, as a matter of fact, not appreciated by society to the same degree. Yes, visual communication is still needed, but it is reverting to the WordArt DIY craze.

Perhaps the temporary surge of high-quality visual communication wasn’t actually caused by the elevated aesthetic standards of our society as I thought, but rather by the absence of visual-making tools for non-professionals. The gap between MS Word and professional graphic-design software was HUGE. Although the browser-based visual design platforms such as Canva filled this gap with editable pre-designed templates and simpler user interfaces, they still did not threaten the existence of graphic designers. But now, with AI embedded directly in Canva, Canva MCP connectors plugged into AI chatbots and AI agents crunching brand manuals in Claude Design - it seems the times of privileged access to professional-looking visual communication tools are over.

Those whose positioning relies solely on the demand caused by limited access to the means of visual production are in real trouble. Even those who spent years perfecting their personal style and developed a unique signature are facing problems nowadays - the more distinctive the visual style, the easier it is to copy it with AI using simple style transfer. Probably the safest position at the moment is that of systems designers: designers who can zoom out from perfecting glyphs and gradients and are drawn to developing holistic design systems. Instead of designing each and every element, they design a system of rules from which the rest is formed automatically. This aligns well with the algorithmic thinking of an AI agent and is a prerequisite for designing AI workflows. Moving to the meta-level of design is one of the survival strategies for designers in times of AI-generated websites, vibe-coded apps and automated creative workflows.

But shape-shifting according to how AI tools evolve and how they change the market can become extremely exhausting. And somewhere along the way you’re probably gonna lose what made you a good designer in the first place. The problem is that most of us are not really sure what it is. It took me a lot of effort and time to figure it out. But on that topic later :)

I spent years assuming that my creativity would protect me. Now I’m beginning to think that looking for the next skill AI cannot perform is the wrong search entirely. The list will keep changing. The harder question is what remains mine when the market no longer needs the outputs I learned to produce.

—

When I came home, I looked at the animation again. It wasn’t so scary anymore. I was even able to find things I would improve. Oh, and I did! With the option to comment on a specific area of the animation and have Claude change just that area, iterating has actually become fun.

This whole dramatic experience showed me how important it is not to allow the AI spectacle to send my brain into overdrive. It’s funny how my nervous system detected the whole situation as a threat and went into full-blown fight-or-flight mode. It made me wonder how often my body recognizes danger in this way, and why the hell I feel in danger while sitting at home with my computer.

Although I was able to reassess the animation through a different lens after calming myself down outside, I don’t think my body’s reaction was in any way misleading. It was intense enough to make me take a break and question what exactly had triggered me. But I am grateful it was there, because I was reacting to the possibility of gradually becoming irrelevant. Of nobody ever wanting my 120% perfection because good enough is now (almost) free. Of feeling left out of the process of creation. And that’s not a small issue at all.

Usually, it’s so easy to ignore these bodily responses. It does not have to be this dramatic every time. Sometimes it might be just a slight tightening in the stomach or an elevated heart rate. You might even respond with unusual fatigue, which might be part of the same kind of activation - a freeze response rather than an impulse to run away.

As a result, I decided to pay much more attention to how my body reacts to working with AI. Just as signals in the wild direct our attention to something that might endanger our existence, these responses might act as a smart detection mechanism for moments when our agency is at risk.

I won’t judge the severity of the reaction because, at this point, it is impossible to assess how endangered our cognitive agency within AI systems really is. But I will keep an eye on it, because the moment my body reacts, I know that something important is happening and that it is worth exploring.

—

I’m currently working on a book about what AI is doing to us - the new habits, strange behaviors and small moments when we suddenly notice ourselves changing. This essay is part of that thinking in public, so if it made you recognize something in your own life, I’d love to hear about it.